CONSUMPTION AND ASSET ALLOCATION WITH INFORMATION LEARNING AND CAPITAL GAINS TAX

In this study, we explore the consumption and asset allocation problem with random horizon in which stock returns are unobservable and capital gains are taxed. The agent with recursive utility could apply the historical stock prices and her private information to learn the expected returns on the stock. We obtain a semi-analytical solution to this problem. Numerical results show that the weights in the non-concealed risky asset and the safe asset increases and decreases with the expected return learnt, respectively. Besides, as the accuracy of estimation gets improved, the agent will reduce her consumption and riskless asset investment and enhance her investment in the non-concealed risky asset. However, information learning does not affect the tax evasion strategy. We also find that the effect of risk aversion on consumption-wealth ratio is significantly affected by the estimation accuracy.